Introduction Colorectal cancer (CRC) remains a leading cause of mortality worldwide. The incidence has increased by 6% in the last decade and it is the third most common malignancy in the UK, accounting for approximately 15,000 deaths annually. The UK Bowel Cancer Screening Programme has demonstrated that detection of CRC at an earlier stage and identification of advanced pre-malignant adenomas can reduce cancer-associated mortality. We analysed the volatile organic compounds (VOCs) emitted from stool to assess their utility as a biomarker for colonic neoplasia.

Methods Stool was collected from symptomatic patients referred for colonoscopy and those within the BCSP. SPME headspace extraction followed by GC-MS was performed, in order to identify VOCs. The outcome of colonoscopy was classified as no-neoplasia, adenomatous polyp(s) and cancer. Data analysis was performed in R, Stata and Metaboanalyst, utilising Student’s t test, Fisher’s exact test, ANOVA, false discovery rate correction, PLS-DA, factor analysis and ROC analysis. Logistic regression modelling with 10 fold cross validation was used to test potential biomarkers.

Results 137 patients were included, with a mean age of 64 yrs (range 22–85), 54% were males. 60 patients had no neoplasia, 56 adenomatous polyp/s and 21 had CRC. 162 VOCs were identified across all samples, mean number in no-neoplasia 58.1 and 55.2 in neoplastic group, p = 0.51. VOC analysis was able to differentiate those with higher risk neoplastic disease with the greatest confidence. When comparing those with cancer to no neoplasia PLS-DA demonstrated separation, compound A was significantly more abundant in the CRC samples (p =< 0.0001, q = 0.004), AUROC curve 0.76. When combined with compound B the AUROC curve was 0.82. Biomarker modelling, combining compounds A and B, used logistic regression and 10 fold cross validation: the AUROC for the initial cross validation set was 0.85, when applied to the hold-out set the AUROC was 0.82, sensitivity 87.9% (95% CI 0.87–0.99) and specificity 84.6% (95% CI 0.65–1.0). Further logisitic regression analysis of VOC presence identified a three VOC panel (compounds A, X and Y) AUROC =0.86: a person 6 times more likely to have cancer for each of the VOCs present in their stool (p =< 0.0001). Factor analysis supported these findings.

*VOCs referred to as Compound A, B, X and Y due to potential future IP.

Conclusion VOC analysis has a superior diagnostic ability for the identification of colorectal adenocarcinoma, when compared to other faecal based biomarker, including those currently employed in UK population based screening.

Disclosure of Interest None Declared